Search Results (37,695)

This study analyzes the transition toward anticipatory public administration through the use of artificial intelligence (AI). Despite the growing deployment of predictive models, a clear research gap remains regarding the socio-technical prerequisites—such as institutional trust, data infrastructure, and ethical–legal frameworks—that condition their success. To address this, a Systematic Literature Review (SLR) was conducted following the PRISMA 2020 guidelines, retrieving 68 open-access articles published between 2020 and 2025 from the Scopus and Web of Science databases. Data synthesis was performed using descriptive bibliometric mapping and qualitative thematic analysis. The results show that scientific interest has experienced accelerated growth since 2024, highlighting a preference for machine learning systems that automate service delivery. While AI enables significant benefits, including a 40% reduction in budgetary errors and early fraud detection, progress is hindered by algorithmic opacity (“black box” models), the risk of structural bias, and civil servants’ confirmation bias. The study concludes that technological complexity must be subordinated to democratic safeguards. To this end, a research agenda is proposed alongside practical implications, recommending that public institutions implement mandatory independent algorithmic audits, enforce strict interpretability standards in public procurement, and adopt hybrid regulatory frameworks to ensure fair and accountable governance. Full article

Bridge health monitoring and machine learning are increasingly intertwined for civil engineers and artificial intelligence experts. Bridges’ poor health can result in severe outcomes if not addressed in time. Therefore, continuous monitoring is required to detect any anomaly or damage. Sensors, such as accelerometers, inclinometers, thermistors, etc., can help actively monitor these bridges. The signals from these sensors help record physiological activities. Such activities are helpful for anomaly detection, damage localization, and bridge health predictions with the help of machine learning algorithms. The proposed method extracts features from the dynamic response of a bridge to ambient excitation. It focuses on processing the signal received from different accelerometers installed on a steel railway bridge to determine the location of the damage and the level of the damage predictions. Initially, features are extracted from time-series data; then, they are fed to a deep neural network after some pre-processing. Normal and augmented data are used with different parameter tuning for results. Original data is also subdivided, and the effect of data slicing on the predictions is investigated. The results show that one-fourth of the slicing of the original data gives the best results for training and testing accuracy with a deep neural network. The results show that the reduced matrix representation, particularly the 40 × 40 feature slicing, improved the classification performance for the predefined bridge scenario classes under the considered experimental settings. For bridge scenario classification, the best reported accuracy was 93.54%, while for damage intensity classification the best reported accuracy was 98.21%. In the DNN-based optimizer comparison, the Adam optimizer achieved higher and more stable performance than Stochastic Gradient Descent (SGD), with test accuracies of 92.3% and 93.7% compared with 75.2% and 86.4%, respectively. It is also observed that the Adam optimizer outperformed Stochastic Gradient Descent (SGD) in terms of both damage localization and damage intensity estimation. Full article

With an ever-growing recognition of Land Surface Temperature (LST) as a key Essential Climate Variable (ECV), it becomes utmost important to have such a variable at the fine spatial and temporal scales of urban spaces and dynamics. Sentinel-3 provides coarse LST (1 km, daily) based on thermal imagery acquired by its Sea and Land Surface Temperature Radiometer (SLSTR) as well as fine Spectral Directional Reflectances (SDRs, 300 m, every two days) synergically inferred from both SLSTR and the Ocean and Land Colour Instrument (OLCI), which gives the opportunity for using the latter as a predictor in the downscaling of the former. Herein, two scale-invariance-based architectures were developed: a single-timestamp (STS) model, trained with coarse data of the timestamp whose fine target it infers; and a multi-timestamp (MTS) one, trained with multiple timestamps. Note that while several Machine Learning (ML) models besides Linear Regression (LR) were considered for the MTS architecture, only LR was used for the STS one due to the limited amount of available data which the former require for hyperparameter tuning. The models were developed over four Danish Functional Urban Areas (FUAs) using SRD-derived indices and seasonal and geospatial predictors and validated against Landsat data. While Gradient Boosting (GB) achieved the best coarse-scale performance it corresponded to the worst fine-scale performer together with Random Forest (RF), indicating scale invariance breakdown. Tree-based models performed poorly due to extrapolation limitations, whereas Neural Net (NN) and LR proved more robust. After residual correction, single-timestamp LR achieved the best fine-scale performance, making it the most reliable and recommended architecture for operations. The overall results showed that, although ML models may better predict the target at their training scale, their performance may not significantly generalise at others, therefore revealing scale specificity. Furthermore, the results suggested that usage of the more general multi-timestamp architecture instead of the single one may deteriorate performance. Full article

Landfills represent a significant source of methane emissions, with important environmental, climatic and safety impacts due to the widespread and variable nature of these emissions. Traditional monitoring methods, such as flow chambers coupled with flame ionisation detectors (FIDs), provide high accuracy but are limited in terms of spatial representativeness, operational flexibility and cost, especially during large-scale or continuous monitoring campaigns. Within this context, the European ESCAPE project aims to develop a low-cost, portable and modular platform for the detection and quantification of low methane concentrations in ambient air at complex environmental sites. The system is based on commercial MOX and NDIR sensors integrated into portable toolboxes equipped with dedicated chambers, regulated suction systems and autonomous data acquisition units with real-time transmission. This work describes the development and testing of two identical toolboxes to assess system reproducibility and the transferability of predictive models between devices. Laboratory and field tests were carried out under controlled and real landfill conditions, with comparisons against portable FID measurements. Results showed good agreement between predicted methane concentrations and reference data, with correlation indexes up to 0.77. Moreover, transferring the machine learning model between toolboxes did not produce statistically significant performance reductions, demonstrating promising robustness and generalizability of the proposed calibration strategy. Full article

Full-vehicle crash simulations that account for occupant injury are essential for automobile safety assessment; however, they are computationally intensive and time-consuming. In particular, dash panel deformation plays a key role in transmitting impact loads to an occupant’s lower extremities. To address this issue, this study proposes a two-stage machine learning framework for occupant lower-limb injury assessment. In the first stage, the deformation behavior of the dash panel is predicted using a machine learning model, enabling efficient generation of a wide range of deformation patterns. In the second stage, occupant lower-limb injury metrics are evaluated based on the predicted deformation using a sled model. While the ultimate objective is to establish the complete two-stage framework, the present paper is limited to the first stage. It investigates the feasibility of machine learning-based deformation prediction. Deformation distributions of simplified structural components are predicted using an XGBoost-based machine learning model, in which principal component scores derived from geometric and deformation data serve as input features. The objective is to efficiently generate representative deformation modes from limited training data rather than optimizing prediction accuracy for individual deformation responses. Numerical experiments are conducted to investigate the effectiveness of the proposed prediction framework. The results of the proposed approach show good agreement with crash simulations in overall deformation behavior, while local deformation is not reproduced perfectly. These findings demonstrate the feasibility of machine learning-based dash panel deformation prediction as the first step toward the proposed two-stage framework for lower-limb injury assessment. Full article

Infertility affects millions worldwide, with estimates indicating that 1 in 6 people of reproductive age will experience it in their lifetime. Globally, infertility impacts between 12.6–17.5% of couples of reproductive age. Recently, machine learning (ML) has garnered significant attention in biomedical research, enabling creation of predictive models that can personalize disease treatment based on measurable variables, thereby aiding in the development of diagnostic tools. In this study, 28 predictor variables were selected preliminarily; after a multicollinearity test, 20 predictors were selected for the classification task and modelled as a binary supervised classification problem. Seven ML algorithms were evaluated, including Logistic Regression, Random Forest, Decision Tree, Support Vector Machine, Naïve Bayes, K-Nearest Neighbour, and Extreme Gradient Boosting (XGBoost). Statistical analysis showed that anti-Müllerian hormone (AMH) can serve as a biomarker for diagnosing PCOS and evaluating ovarian reserve. Female fertility has been associated negatively with waist circumference (r = –0.35), systolic blood pressure (r = –0.30), poor ovarian reserve (r = –0.28), and triglycerides (r = –0.33), suggesting a possible link between these metabolic factors and female infertility. Among the models tested, Naïve Bayes and Logistic Regression provided the most reliable and generalizable performance. The incorporation of SHapley Additive exPlanations (SHAP) analysis enhanced the interpretability of the models, identifying polyendocrine metabolic ovarian syndrome (PMOS, previously known as polycystic ovarian syndrome—PCOS), AMH, poor ovarian reserve, menstrual cycle irregularity, systolic blood pressure, body mass index (BMI), fasting glucose, and triglycerides as the most influential predictors of female fertility. However, future studies incorporating data from multiple centres, comprising a larger, more representative population, and using more interpretable models could enhance the reliability of ML in clinical decision-making. Full article

This work presents an interpretable machine learning (ML) system that uses composition- and physics-based descriptors to predict the bulk moduli of high-entropy alloys (HEAs). Extra Trees, Random Forest, Gradient Boosting, AdaBoost, and LightGBM are five ensemble ML algorithms that were systematically shaped and refined by hyperparameter fine-tuning. With a test R² of about 0.852 and an RMSE and MAE of about 5.49 GPa and 1.5 GPa, respectively, Extra Tree outperformed the other optimized models, indicating good generalization capacity for untested HEA compositions. The computational efficiency results showed that LightGBM had the fastest prediction speed (~4.24 ms), whereas Extra Trees had the shortest training time (~17.3 s). The majority of the optimized models had statistically equal prediction performance (p > 0.05), according to statistical validation using paired t-test analysis, even though residual error distributions for the Extra Tree model established consistent and unbiased predictions. To enhance the interpretability of the model, SHAP-based explainable analysis was performed, which included SHAP importance, dependence, and waterfall plots. The SHAP results revealed that the primary determinants impacting bulk modulus behavior in HEAs were Zr content, mean electronegativity, Al content, bond strength, and melting-temperature-related parameters. The proposed framework enables the rapid identification and design of next-generation HEAs by permitting precise and computationally efficient bulk modulus prediction, as well as physically significant insights into descriptor–property connections. Full article

Autonomous driving systems rely on a sophisticated pipeline of artificial intelligence models to perceive, predict, and plan in dynamic environments. This review presents a systematic analysis of the machine learning and deep learning models underpinning vehicle autonomy, spanning classical convolutional neural networks (CNNs) for object detection and semantic segmentation to recurrent and Transformer-based architectures for trajectory prediction and motion planning. It also provides a critical examination of the autonomous vehicle sensor stack, including cameras, LiDAR, radar, ultrasonics, and GNSS/IMU as data acquisition systems, highlighting modality-specific AI challenges such as monocular depth estimation, 3D point cloud processing, and radar Doppler interpretation. The evolution of perception and decision-making pipelines is reviewed, contrasting modular architectures with end-to-end learning paradigms that directly map raw sensor data to control commands, and discussing their trade-offs in interpretability, safety assurance, and robustness to rare edge cases. We further survey specialized hardware accelerators and heterogeneous automotive SoCs designed to meet stringent real-time and power constraints. Industrial strategies are compared, including multi-modal sensor fusion and vision-centric approaches based on large-scale imitation learning. Finally, we identify open challenges related to robustness under adverse conditions, domain shift, causal ambiguity, and the need for interpretable and certifiable AI in safety-critical autonomous driving systems. Full article

Air pollution from particulate matter (PM2.5 and PM10) poses a serious public health risk in urban environments, particularly in areas with heavy vehicular traffic. Against this backdrop, the present study proposes an Internet of Things (IoT)-based system designed to support air quality monitoring and evidence-based decision-making regarding PM2.5 and PM10 concentrations, integrating low-cost sensors with a machine learning prediction module. The study follows an experimental-applied design with a quantitative–comparative approach. Its scientific contribution is organized around an integrated IoT-ML framework addressing a concrete gap in the literature: the lack of local empirical evidence regarding which family of machine learning algorithms delivers the greatest accuracy, stability, and computational efficiency for particulate matter forecasting in mid-altitude urban environments using low-cost sensors. On one hand, the framework proposes and deploys a four-node IoT network for continuous PM2.5 and PM10 monitoring in high-traffic urban microenvironments—representing one of the first sustained deployments with low-cost, high-temporal-resolution sensors (10-minute intervals) in Arequipa, Peru. On the other hand, the study presents the most extensive benchmarking reported in the local literature: a systematic evaluation of 44 machine learning algorithms under homogeneous experimental conditions, covering classical statistical models, traditional machine learning techniques, deep learning architectures, and hybrid approaches, along with an analysis of ensemble learning strategies using Ridge stacking and K-Fold cross-validation. This unified comparative analysis—applying consistent metrics (MAE, RMSE, R², and MAPE), the same prediction horizon, and a shared dataset—provides replicable empirical evidence that had not previously been reported for the urban context of Arequipa. The results show that traditional statistical models perform poorly overall, while tree-based and boosting algorithms consistently achieve R² values above 0.90 for both pollutants. Ensemble models, particularly stacking with Ridge regression and cross-validation, yielded the strongest overall performance, demonstrating greater robustness and prediction stability. Explainability criteria were also incorporated, enabling an assessment of each base model’s individual contribution and identifying the variables most relevant to the prediction process. The methodological contribution provides future researchers with a rigorous reference framework for algorithm selection in environmental IoT systems. Taken together, the findings demonstrate that combining low-cost IoT networks with advanced machine learning and ensemble learning techniques constitutes an effective, scalable, and cost-efficient alternative for air quality monitoring, predictive analysis, and the support of informed mitigation strategies in urban environments. Full article

Physical climate hazards—floods, storms, heatwaves, and wildfires—are increasingly disrupting banking operations and generating growing litigation and legal-risk exposures, yet operational risk remains one of the least studied channels through which climate change may affect financial institutions. This paper provides a critical review of the emerging literature on climate-related operational risk in banking, covering both physical disruptions and the legal risk dimension explicitly recognised within the Basel operational risk framework. We map the empirical evidence, critically evaluate the methodological toolkit—event studies, fixed-effects regressions, difference-in-differences, dynamic panel estimators, and logit models—and assess their suitability for a domain characterised by data scarcity, rare events, and non-linearity. Building on this assessment, we outline a conceptual methodological roadmap intended to guide future research, organised around three stages: (i) machine learning-based variable selection and anomaly detection applied to operational loss and climate databases; (ii) econometric modelling of climate-related operational events with explicit identification strategies; and (iii) agent-based modelling to simulate system-wide propagation of climate shocks. Each stage can be conceptually related to elements of the Basel operational risk framework, offering a structured research programme for academics and a diagnostic toolkit for supervisors and risk managers. Full article

This article presents a study that applies Laban’s Effort theory to detect a musician’s expressive intentions during performance. Laban Effort theory was chosen for its capacity to support the observation, description, and interpretation of expressive movement. The aim was to develop interactive musical systems that learn from the embodied expressivity musicians cultivate through practice. The study adopts a multidisciplinary approach, combining Laban Motion Analysis with interactive machine learning. Gesture data, audio, and video were recorded during a performance of Brahms’s Clarinet Sonata Op. 120 No. 2. The video was then analysed by an expert annotator observing Laban Efforts. The annotated video was then reviewed in collaboration with the performer to incorporate Effort Phrasing that reflects changes in intensity within Efforts. The annotations were then used as training data, together with motion data recorded during the performance, to train a regression model. The model was evaluated against expert annotations and through qualitative video analysis. Unlike the very linear notation of a Laban analyst, the model reflects the dynamism of human motion by capturing the almost humanly unobservable nuances in Effort variations. This points to the possibility of using machine learning models to reflect a performer’s expressive intentions in real-time. Full article

This study investigates the effectiveness of common thermal, climate, and envelope features in predicting annual site energy use intensity (site EUI) for different types of residential buildings in the USA. A proposed multi-level data approach that consists of regression algorithms and feature analysis has been implemented to derive models from different sets of features related to thermal, envelope, and climate, respectively. Feature set analysis is conducted using correlation analysis methods besides chi-square testing (CHI) and gain ratio (GR) methods to offer interpretable global features rankings. Models were developed using regression-based algorithms (linear, lasso, and ridge) under a 10-fold cross-validation on different distinct sets of features besides permutation feature importance (PFI) analyses to validate the models in terms of root mean squared error (RMSE). The novelty of this study lies in the comparison of feature groups and the evaluation of their individual and incremental contributions to site EUI prediction. Results against the WiDS Datathon 2022 building energy dataset demonstrate consistently ranked climate and thermal indicators (accumulated annual heating degree days (AAH) and accumulated annual cooling degree days (AAC), and heating dominance (HD), cooling dominance (CD), snowfall, and extreme temperature days) as the most informative predictors among the evaluated feature groups. The model with the best performance has an RMSE value of about 38.68; however, from the low Coefficient of determination (R²) values, it can be noted that yearly climatic conditions and building envelope characteristics cannot be only used to account for the variation in site EUIs on their own, thus showing the need to consider other factors. Full article

Groundwater-potential assessment in karst aquifers is complicated by pronounced spatial heterogeneity driven by structural permeability, lithological variability, recharge redistribution, and unresolved subsurface conduit connectivity. Although machine-learning approaches have improved regional groundwater mapping, most existing models provide only deterministic predictions and offer limited information on predictive uncertainty and hydrogeological reliability. To address this limitation, we propose CT-TreeFlow. This probabilistic groundwater assessment framework goes beyond conventional machine-learning models by explicitly learning the full conditional probability distribution of groundwater favourability rather than a single deterministic estimate. The framework integrates sparse probabilistic environmental routing, conditional density estimation, hydrogeologically constrained pseudo-absence generation, geographically structured spatial validation, and explainability-driven interpretation within a unified modelling architecture, enabling simultaneous groundwater prediction, uncertainty quantification, and hydrogeological interpretation. The framework was applied to the Zagros karst system in Khuzestan Province, Iran, using remote-sensing-derived environmental predictors, Copernicus DEM-based morphometric variables, geological–structural datasets, and hydroclimatic indicators. Performance was evaluated against LightGBM and XGBoost using GroupKFold spatial cross-validation. CT-TreeFlow achieved a mean RMSE of 2.737 and a mean R² of 0.852, while also providing spatially explicit uncertainty estimates and probabilistic prediction intervals. Explainability analyses identified fracture density, lithology, drainage organisation, and terrain-controlled recharge conditions as the dominant controls on groundwater favourability. Predicted high-favourability zones showed strong spatial correspondence with major carbonate formations and independent spring–cave inventories, supporting the hydrogeological plausibility of the mapped patterns. These results demonstrate that probabilistic modelling can provide more reliable and physically interpretable groundwater assessments than deterministic approaches in structurally complex karst environments. CT-TreeFlow offers a transferable framework for uncertainty-aware groundwater exploration and regional hydrogeological decision support in heterogeneous aquifer systems. Full article

To control the quality of mineral-medicinal waters and ensure their therapeutic benefits, spas often rely on periodic discrete sampling to analyze the physico-chemical properties of their pools. The AQUAPRED project aims to digitize this process by deploying IoT systems within the spa facilities, enabling real-time data acquisition via calibrated multi-parameter probes. Using data collected by these pilot systems, we develop and validate a predictive machine learning model capable of forecasting the short-term evolution of the thermal water properties. Historical data from each facility allow the model to learn the specifics dynamics of each spa. As a practical application, we propose an anomaly detection module based on residual analysis from predicted and observed values. Significant discrepancies signal events of interest and emergent trends, such as anomalous readings, contamination or sensor drift. The methodology is evaluated using real data from six spas associated with the AQUAPRED project. The results demonstrate the model’s effectiveness and support its feasibility for deployment in other thermal establishments. Full article

Rapid urbanization has significantly altered urban landscape composition and configuration, making it a key driver exacerbating the urban heat island (UHI) effect. As a rapidly expanding inland city in Central China, Zhengzhou is highly sensitive to changes in landscape composition and spatial configuration. Therefore, clarifying the nonlinear relationship between landscape patterns and the urban thermal environment is of great significance for sustainable urban planning and thermal environment regulation. Taking the main urban area of Zhengzhou as the study area, this paper retrieves land surface temperature (LST) using the radiative transfer equation method based on Landsat 8 remote sensing images from August 2015 to August 2024, and constructs the surface urban heat island intensity (SUHII) index. By integrating multi-dimensional landscape pattern indices, the XGBoost machine learning model, and the SHAP interpretability method, this study systematically analyzes the nonlinear response mechanisms of landscape composition and configuration to SUHII, key regulatory thresholds, and their changes between 2015 and 2024. The results show that: (1) The SUHII in Zhengzhou was substantially higher in 2024 than in 2015. The area proportions of strong and extremely strong heat islands were higher in 2024 (26.16% and 2.34%) than in 2015 (2.22% and 0.12%), and the thermal environment differed between 2015 and 2024, shifting from a localized patch pattern to a more continuously expanding pattern. (2) Landscape area-related indices are the key factors. The areas of green space and water bodies, along with the landscape diversity index, show significant negative correlations, while built-up area and aggregation index show significant positive correlations. (3) SHAP feature importance indicates that water body area is the primary cooling factor, whereas built-up area is the primary warming factor, jointly dominating the spatial pattern of the thermal environment in Zhengzhou. (4) Landscape composition and configuration exhibit significant nonlinear responses to SUHII with region-specific thresholds, and these thresholds were higher/lower in 2024 than in 2015, suggesting a possible association with urban expansion. Specifically, stable cooling effects occurred when the water body area exceeded 3.5 km² in 2015, with the threshold rising to 4.2 km² in 2024. The warming threshold for built-up area decreased from 18.8 km² to 8.5 km², suggesting a higher sensitivity of the thermal environment to built-up area expansion in 2024 compared to 2015, characterized by a regulation pattern of “dominant scale effect and weakened configuration effect”. This study identifies thresholds specific to Zhengzhou’s main urban area at two time points (2015 and 2024), providing quantitative support and scientific basis for blue–green space optimization, precise heat island mitigation, and territorial spatial planning in Zhengzhou. These findings are based on a comparison of two time points (2015 and 2024) and do not directly capture continuous temporal dynamics. Full article